[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Fri, 16 Dec 2011 13:03:48 -0600
From: "jfoug" <jfoug@....net>
To: <john-dev@...ts.openwall.com>
Subject: RE: optimized Lotus5
>> >Please help me test it (run the test suite, etc.) I only tested with
>> >--test so far.
>
>I was referring to the new Lotus5 code in 1.7.9-jumbo-4.
Well, I am fully testing ALL builds, and all formats.
Here are the first 2 runs. 32 bit SSE2 and 32 bit SSE2i, straight from
179-j4 sources
$ ./jtrts.pl -b ../../john-1.7.9/john-1.7.9-jumbo-4/run -t full -q q
----------------------------------------------------------------------------
---
- JtRTestSuite. Version 1.10α1, Dec 15, 2011. By, Jim Fougeron
- Testing: John the Ripper password cracker, ver: 1.7.9-jumbo-4
[win32-cygwin-x86-sse2i]
----------------------------------------------------------------------------
----
All tests passed without error. Performed 210 tests. Time used was 992
seconds
$ ./jtrts.pl -b ../../john-1.7.9/john-1.7.9-jumbo-4/run -t full -q q
----------------------------------------------------------------------------
---
- JtRTestSuite. Version 1.10α1, Dec 15, 2011. By, Jim Fougeron
- Testing: John the Ripper password cracker, ver: 1.7.9-jumbo-4
[win32-cygwin-x86-sse2]
----------------------------------------------------------------------------
----
All tests passed without error. Performed 210 tests. Time used was 1022
seconds
Thus, all -test=0's for all encodings were fine. Also all formats, and test
input files, and methods of testing, showed exactly what they were supposed
to for these 2 builds.
I will keep building and testing. I did notice several sections within
dynamic_fmt.c where the jumbo did not have proper indenting, compared
against what I had been working on. Likely the patching process failed a
little. These are totally non-issues, but I will get them updated next time
I update the dynamic format, which I am actively working on.
However, from these first 2 tests, things look good. I will finish up the
32 bit, then move this code over to 64 bit sparc cc build, and run the test
suite there. That will also test the test suite some more, lol.
The test suite is now a perl script, and a 'data' file. The data file is a
CSV file, where each line describes a test, contains attributes for the test
(which can be selected or forced non-selection), contains other information
on certain requirements that have to be met for this test line to run. So
now, we have a script, data, and then the user can specify a sweeping set of
rules, or be very specific to where only 1 line of the test data is run, or
anywhere in between. The new test suite should be pretty close, but I still
have to make sure that it is FULLY properly handling all cases where there
are problems. It works like a champ when things are good. It also handles
all of the problems I have seen so far (and is far better at listing what
went wrong). It also will never perform tests on formats NOT built into
this build of john. It knows how to handle things differently between core
and jumbo john. It gets all encoding types from -enc:LIST. It gets all
dynamic types from -sub:LIST, and gets all format types from ./john with no
params. It also gets many other things from that ./john without params
call.
It is 'close', and will likely be released this weekend. I would like to
get it compatible all the way back to 1.7. The differences here, being some
params were added, so there may be times where even though it is a jumbo
build, I have to use 'core' type logic. Also, -enc would not be there,
fmt_dynamic was fmt_md5_gen, etc. So there would be some internal
conversion of data to get things working in older versions. However, these
changes will probably not be in the initial 1.10 release. That will simply
be working fully for 1.7.9 +
Jim.
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
